J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003. Genetic Inferences and Breeding Implications from Analysis of Cranberry Germplasm Profi les

Nicholi Vorsa1 and James Polashock2 Philip E. Marucci Center for Blueberry and Cranberry Research and Extension, Rutgers University, Chatsworth, NJ 08019 David Cunningham3 and Robin Roderick4 Ocean Spray Cranberries, Inc., One Ocean Spray Drive, Lakeville-Middleboro, MA 02349

ADDITIONAL INDEX WORDS. Vaccinium macrocarpon, fruit, methoxylation, glycosylation, ,

ABSTRACT. A diversity of exists among angiosperm species. Studies indicate that various anthocyanins differ in antioxidant potential, their bioavailability, and stability during processing. The fruit of the American cranberry, Vaccinium macrocarpon Ait., is recognized as having six anthocyanins, composed largely of 3-O-galactosides and 3-O-arabinosides, and to lesser amount (≈6%), 3-O-glucosides of the aglycones cyanidin and peonidin. This study analyzed proportions of these six anthocyanins from >250 accessions of a germplasm collection over harvest dates. Fruit samples from 78 selected accessions, based on the fi rst year analysis, were also analyzed a second year. Principal component analysis identifi ed general negative relationships between the proportions of cyanidin versus peonidin, arabinosides versus glucosides, and galactosides versus arabinosides and glucosides. These relationships were consistent across the 2 years. Most variation in germplasm anthocyanin profi les refl ected variation of cyanidin versus peonidin proportions, with cyanidin to peonidin ratios ranging from 3.6:1 to 0.5:1. Variation for glycosylation profi les was also evident, with galactoside proportions ranging from 64% to 75%, arabinoside proportions ranging from 20% to 33%, and glucoside proportions ranging from 3% to 9%. Evidence for both signifi cant qualitative and quantitative genetic variation exists for the methoxylation of cyanidin to peonidin. Signifi cant quantitative genetic variation is also apparent for glycosylation. Signifi cant differences exist among anthocyanins as to their Wrolstad, 1990). Although total anthocyanin content (TAcy) of antioxidant potential (Cao et al., 1996). Antioxidant potential ap- fruit is one of the major traits evaluated in cranberry breeding pears to be a function of both the glycoside, as well as the specifi c (Sapers et al., 1983a), information on variation of proportions aglycone, i.e. cyanidin, , pelargonin, etc. (Satue-Gracia of individual anthocyanins is generally lacking. Earlier studies et al., 1997; Wang et al., 1997). For example, Wang et al. (1997) concluded that there were either no or very little qualitative or found cyanidin-3-glucoside to have ≈75% greater antioxidant quantitative differences in the proportions of individual antho- activity than cyanidin-3-galactoside. Cyanidin-3-glucoside was cyanins among the cranberry varieties or harvest dates studied found to have >90% greater antioxidant activity than peonidin- (Sapers et al.,1983b; Sapers and Hargrave, 1987). These studies 3-glucoside (Wang et al., 1997). Evidence is also increasing for did not consider the proportions of glucosides, most likely assum- the signifi cant role that the carbohydrate moiety plays in the ing they were of little consequence. The proportion of cyanidin to absorption of numerous fl avonoids in the gastro-intestinal tract peonidin anthocyanins was determined to be about equal (Hong (Gee et al., 2000; Hollman and Katan, 1997; Hollman et al., 1997; and Wrolstad, 1986, 1990; Sapers and Hargrave, 1987), however, Setchell et al., 2001). Glucoside conjugates appear to be the most these studies evaluated only a limited number of varieties. bioavailable. Hollman and Katan (1997) found that quercetin- We have assembled a large collection of cranberry germplasm 3-glucoside was absorbed to a signifi cantly greater degree than accessions from cultivated beds. Analyzing anthocyanin profi les quercetin-3-rutinoside in human subjects. Mizuma et al. (1994) of a broad array of germplasm over harvest dates and years may in studies with rat intestines found fl avonols as glucosides were provide insight into the genetic variation and inter-relationships absorbed to a greater degree than the galactosides. Anthocyanins among the specifi c anthocyanins in cranberry, as well as providing also differ in their stability during processing (Wrolstad et al., useful information for breeding of desirable anthocyanin profi les. 1994). The objectives of this study were to 1) analyze the proportions The American cranberry (Vaccinium macrocarpon Ait.) is of the individual anthocyanins in a broad array of germplasm recognized for its brilliant red fruit, which is a result of a fairly obtained from cultivated beds over harvest dates and years to high abundance of anthocyanins in the fruit epidermal tissues. The better assess the variation for anthocyanin profi les in American anthocyanins of V. macrocarpon are mainly the 3-O-glycosides cranberry, 2) identify relationships, or possible tradeoffs, among of cyanidin and peonidin. Galactosides and arabinosides of cy- the anthocyanins using principal component analysis, and 3) iden- anidin and peonidin have been reported to be the most abundant, tify varieties with non-typical anthocyanin profi les. with lesser (≈6%) amounts of glucoside anthocyanins (Hong and Materials and Methods Received for publication 19 Oct. 2002. Accepted for publication 18 Feb. 2003. We thank Kate Brilliant and Sarah Vannozzi for their technical assistance, and Joseph PLANT MATERIAL. Germplasm used in this evaluation consisted DeVerna, for helpful suggestions and critically reviewing the manuscript. of 252 clonal accessions collected between 1988 and 1993 from 1Professor; corresponding author. 2Assistant research director. Current address: USDA–ARS–Fruit Lab, Chatsworth, cultivated beds in Massachusetts, New Jersey, Oregon, Wash- NJ 08019. ington, and Wisconsin, which were established in 2.3-m2 fi eld 3Principal scientist. plots. Germplasm plots were located in one of two 0.25-ha beds 4Senior scientist. (Beds 1 and 4) at the Rutgers (N.J.) Univ. P.E. Marucci Center.

J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003. 691

9121-Genet 691 7/10/03, 10:59:04 AM Bed 1 has a 25-cm-deep washed sand base, while Bed 4 consists acid as a standard. The HPLC was a binary pump system (Hewlett of berryland soil. Packard 1100) with a variable diode-array detector. UV-visible FRUIT SAMPLES, TACY AND HPLC ANALYSIS. Fruit samples from spectra were recorded at 280, 320, and 520 nm. The column was 252 plots were harvested from 930-cm2 areas (minimally 30 cm from a Prodigy ODS-3, 5 µm, 250 × 4.6 mm, protected with a Prodigy plot edge) from each plot on 5 Sept. and 5 Oct. 1998. In 1999, 78 guard ODS-3, 5 µm, 30 × 4.6 mm. Separations were carried out accessions were selected (accessions were selected to represent the using the following conditions: 1 mL·min–1 fl ow rate, 25-µL injec- full range of variation based on 1998 principal component analysis) tion volume; column compartment temperature 30 °C; solvent A, for fruit samples on 20 Sept. and 11 Oct. The data obtained for each 2% (v/v) acetic acid in water; solvent B, 80% (v/v) acetonitrile and fruit sample harvested were total weight, berry count, mean fruit 20% solvent A; elution with linear gradients from 0% B (balance weight, damaged fruit weight and number and chemical analyses 100% A) for 3 min, from 0% B to 4% B in 3 min, from 4% B to are described below. 10% B in 9 min, from 10% B to 15% B in 15 min, from 15% B to For both years, fruit samples (50 g) were frozen at –20 °C. The 23% B in 20 min, from 23%B to 25% in 10 min, from 25% B to samples were analyzed randomly across plots (accessions) and 30% B in 6 min, from 30% B to 50% B in 14 min, from 50% B to harvest dates. The 50 g frozen fruit sample along with 60 mL of 80% B in 3 min, and then back to 0% B in 2 min, followed by a hot (50 °C) tap water were placed in a blender and macerated for 3 20 min post time. Individual anthocyanins were quantifi ed using min. From the resultant slurry, the total anthocyanins (TAcy) were a standard of cyanidin-3-galactoside at 520 nm. TAcy-HPLC was determined using a standard spectrophotometric assay adapted from calculated as the sum of areas representing the peaks at retention Sapers et al. (1983b). Concurrently, the sample was also prepared time (RT) for the six anthocyanins. for analysis of total phenolics, individual anthocyanins and total STATISTICAL ANALYSIS. Statistical analysis employed SAS 8.0 anthocyanins by HPLC (TAcy-HPLC). The slurry was centrifuged for Microsoft Windows, SAS Institute, Inc., Cary, N.C. Procedures (40,790 ×g, 10 min, 25 °C), the supernatant fi ltered (0.45 µm), the used were PROC CORR for Pearsonʼs correlation coeffi cients, fi ltrate sealed in a 2 mL amber glass injection vial, and frozen until PROC PRIN for principal component analysis, and PROC GLM analysis. The HPLC method used was adapted from Lamuela- for analysis of variance with Type III sums-of-squares for testing Raventos and Waterhouse (1994), substituting catechin for gallic bed, harvest date and variety effects.

Table 1. Eigenvalue percentages for principal components (PC) 1–5 vectors derived from the absolute levels of six cranberry anthocyanins in the germplasm collection. 1998 1999 5 Sept. 5 Oct. Combined 20 Sept. 11 Oct. Combined N = 252 N = 245 N = 497 N = 78 N = 78 N = 156 PC 1 92.0 89.2 91.8 94.7 86.3 91.2 PC 2 5.2 6.1 4.7 3.1 7.5 4.6 PC 3 2.5 4.1 3.1 1.9 5.6 3.8 PC 4 0.2 0.4 0.2 0.2 0.4 0.3 PC 5 0.1 0.2 0.1 0.1 0.1 0.1

Table 2. Pearsonʼs correlation coeffi cients for correlation among levels of the six anthocyanins (cyanidin-3-arabinoside, cyanidin-3-galactoside, cyanidin-3-glucoside, peonidin-3-arabinoside, peonidin-3-galactoside, peonidin-3-glucoside), with one another as absolute values (above diagonal) and as percentages of the total anthocyanins of the sample (below diagonal). Datez Cy-3-gal Cy-3-arab Cy-3-glu Pn-3-gal Pn-3-arab Pn-3-glu Cy-3-gal S5-98 --- 0.98y 0.96 0.91 0.90 0.88 O5-98 --- 0.98 0.95 0.91 0.88 0.87 99 --- 0.97 0.95 0.91 0.89 0.87 Cy-3-arab S5-98 0.72 --- 0.92 0.87 0.89 0.81 O5-98 0.65 --- 0.92 0.89 0.89 0.81 99 0.66 --- 0.86 0.89 0.91 0.80 Cy-3-glu S5-98 0.57 0.13** --- 0.88 0.86 0.91 O5-98 0.66 0.16 --- 0.85 0.80 0.89 99 0.75 0.23** --- 0.85 0.80 0.90 Pn-3-gal S5-98 –0.93 –0.85 –0.50 --- 0.99 0.97 O5-98 –0.91 –0.82 –0.60 --- 0.98 0.95 99 –0.97 –0.87 –0.65 --- 0.98 0.95 Pn-3-arab S5-98 –0.91 –0.31 –0.69 0.76 --- 0.95 O5-98 –0.92 –0.29 –0.70 0.71 --- 0.90 99 –0.93 –0.37 –0.65 0.75 --- 0.90 Pn-3-glu S5-98 –0.56 –0.84 0.25 0.64 0.29 --- O5-98 –0.48 –0.77 0.24 0.54 0.23 --- 99 –0.39 –0.79 0.22** 0.54 0.10NS --- zHarvest dates are S5-98 (5 Sept. 1998), O5-98 (5 Oct. 1998) and 99 (20 Sept. and 11 Oct. 1999). yAll values signifi cant at less than P = 0.0001, unless noted otherwise. NS,**Nonsignifi cant or signifi cant at P = 0.01, respectively.

692 J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003.

9121-Genet 692 7/10/03, 10:59:08 AM Results The weightings in PC1 indicate that there is a general negative correlation between cyanidin versus peonidin glycoside proportions CRANBERRY ANTHOCYANIN PROFILE VARIATION: INTER-RELATION- for fruit harvested both years, although the signs were reversed SHIPS OF INDIVIDUAL ANTHOCYANINS. TAcy-HPLC quantities (the between the years (Table 4). The assignment of sign (positive or six anthocyanins combined) ranged from 9.6 µg·g–1 (US88-85, 5 negative) to variables in PROC PRIN follows the arbitrary rule Sept.) to 1,500 µg·g–1 (US88-119, 5 Oct.) in 1998. In 1999, total that the sum of eigenvectors is greater than zero (Anonymous, SAS anthocyanins ranged from 31.1 µg·g–1 (US 88-92, 20 Sept.) to Institute, Inc.). Except for the low (but signifi cant) positive cor- 1,108.5 µg·g–1 (US88-119, 11 Oct.). Total anthocyanins (TAcy) relation between cyanidin-3-glucoside and peonidin-3-glucoside, ranged from 3 to 97 mg/100 g fruit in 1998 and from <1 to 125 the relationships between the cyanidin versus peonidin glycosides mg/100 g fruit in 1999. TAcy was fairly highly correlated with were all negative, varying in magnitude from r = – 0.29 between TAcy-HPLC values, with r = 0.88 (P < 0.001, N = 497) in 1998 arabinosides, to r = – 0.93 between the galactosides (Table 2, below and r = 0.95 (P < 0.001, N = 164) in 1999. diagonal). PC2 accounted for 27.2% to 32.2% of the total varia- Principal component analysis indicates the majority of the variation tion in anthocyanin proportions (Table 3). The PC2 eigenvectors was due to differences in overall absolute values of the six anthocya- suggest a general negative relationship between arabinosides and nins, resulting from differences between varieties and harvest dates. glucosides. The glucosides of cyanidin and peonidin have positive The fi rst principal component (PC1) accounts for between 86.3% to weightings, while the arabinosides of cyanidin and peonidin are 94.7% of the variation (Table 1), with about equal weightings for the negatively weighted, indicating a negative correlation between the absolute levels of the six anthocyanins for both 1998 and 1999 (data glucosides and arabinosides (Table 4). The galactoside anthocya- not given). Variation arising from differences in relative proportions nins in PC2 were essentially neutral, as indicated with weight- of the six anthocyanins accounted for about 8% of the total variation ings nearer to zero ranging from 0.03 to 0.12. PC3 accounted for (Table 1). Within harvest dates, the variation attributable to differences 3% to 5% of the variation (Table 3). The weightings of the PC3 in proportions of the six anthocyanins ranged from 5.3% (20 Sept. eigenvectors suggest a general negative relationship between the 1999) to 13.7% (11 Oct. 1999) (Table 1). The absolute values of the galactoside proportions versus proportions of both arabinosides individual anthocyanins were all signifi cantly positively correlated and glucosides, and were consistent between the 2 years (Table with each other (Table 2, above diagonal). The highest correlation 4). The galactosides of cyanidin and peonidin had about equal was observed between the galactosides and arabinosides of cyanidin negative weightings, while the weightings for glucosides and ara- (r = 0.98) and peonidin (r = 0.98), respectively. The lowest correla- binosides of both cyanidin and peonidin were positive, indicating tions were cyanidin-3-arabinoside with peonidin-3-glucoside (r = galactosides are negatively associated with that of arabinosides and 0.81) and cyanidin-3-glucoside with peonidin-3-arabinoside (r = glucosides. It appears that although the glucoside and arabinoside 0.86) (Table 2, above diagonal). proportions are still positively correlated with the galactoside pro- To identify relationships among the levels of six anthocyanins, portions within an aglycone class, the correlation values are only the levels of the six anthocyanins were analyzed as proportions moderate, ranging from r = 0.54 to 0.76 (Table 4, below diagonal). of the total anthocyanins within a sample. Relationships among The weightings of the arabinosides appeared to be slightly higher the proportions of the six anthocyanins are evident in the fi rst than for the glucosides. (PC1), second (PC2) and third (PC3) principal components, and ANTHOCYANIN PROFILE VARIATION. A plot of PC1 versus PC2 were essentially consistent across harvest dates (not shown) and displays the variation for cyanidin versus peonidin anthocyanin the two years (Table 3). PC1 accounted for ≈66% of the total proportions (PC1), and arabinose versus glucose anthocyanin variation in the six anthocyanin proportions. proportions (PC2), respectively, in 1998 (Fig. 1). The majority Table 3. Eigenvalue proportions for principal components (PC) 1–4 vectors derived from the proportions of six cranberry anthocyanins in the germplasm collection. 1998 1999 5 Sept. 5 Oct. Combined 20 Sept. 11 Oct. Combined N = 252 N = 245 N = 497 N = 78 N = 78 N = 156 PC 1 66.3 66.1 66.7 66.6 63.6 66.7 PC 2 27.8 27.4 27.2 27.2 32.2 27.6 PC 3 3.9 4.7 4.2 4.8 3.3 4.5 PC 4 1.2 1.5 <1 1.2 0.7 1.1

Table 4. Eigenvectors for principal components 1–3 (PC1, PC2 and PC3) for proportions of six cranberry anthocyanins in cranberry germplasm across two harvest dates for 1998 and 1999. PC1 PC2 PC3 Anthocyanin 1998z 1999y 1998z 1999y 1998z 1999y Cy-3-gal –0.49 0.49 0.12 0.10 –0.26 –0.28 Cy-3-glc –0.26 0.35 0.64 0.50 0.42 0.40 Cy-3-arab –0.41 0.39 –0.39 –0.43 0.49 0.47 Pn-3-gal 0.49 –0.49 0.03 0.09 –0.31 –0.28 Pn-3-glc 0.32 –0.24 0.57 0.65 0.27 0.37 Pn-3-arab 0.43 –0.43 –0.32 –0.33 0.59 0.56 zHarvest dates are 5 Sept. and 5 Oct. yHarvest dates are 20 Sept. and 11 Oct.

J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003. 693

9121-Genet 693 7/10/03, 10:59:12 AM Fig. 1. Plot of the 1998 accession means (across two harvest dates) of fi rst (PC1) and second principal (PC2) components derived from the levels of proportions of the six major cranberry anthocyanins.

of accessions were within the range of – 4 and 4 for PC1, having cyanidin/peonidin proportions of 66/34 (2:1) to 42/58 (0.7:1). One apparent outlier was accession US89-7 which had 78.3% cyanidin (21.7% peonidin) anthocyanins or a 3.6:1 cyanidin:peonidin ratio. There was also another outlier group (US93-246, 247 & 248) located at PC1 > 7, having a relatively low proportion (30%) of cyanidin (70% peonidin) anthocyanins, or a 0.5:1 cyanidin:peonidin ratio. Accessions US93-246, 247 and 248 were found to have identical randomly amplifi ed polymorphic DNA (RAPD) and sequence characterized amplifi ed regions (SCAR) fi ngerprints suggesting Fig. 2. Mean (across accessions) proportions of six cranberry anthocyanins that they are, in fact, identical clones (Novy et al., 1994; Polas- for two harvest dates in 1998 and 1999. hock, unpublished data). The ratio of proportions of arabinoside to glucoside anthocyanins ranged from 30.4:2.8 (US89-10) to proportions of peonidin anthocyanins. Both bed and variety effects 20.4:8.4 (US93-223) arabinose:glucose or ratios of 10:9 to 2.4:1, were signifi cant for PC2 suggesting both environmental and genetic respectively. PC3 was signifi cantly negatively correlated with the effects are signifi cant for arabinoside versus glucoside proportions. proportion of galactoside anthocyanins (r = –0.74). PC3 values Bed, variety and harvest date effects were signifi cant for PC3 indi- ranged from -1.9 to 1.6, with galactoside anthocyanin proportions cating the proportions of galactosides versus arabinoside/glucoside of 63.5 (US92-11) to 77.4 (US88-78) percent, respectively. Cor- have both environmental and genetic effects. relation of the principal component means (by accession) between The ranges of proportions of the six anthocyanins of the com- the two years (78 genotypes) were highly positively signifi cant, mon cultivars Stevens, Early Black, and Ben Lear are 30% to 41% with r = –0.96 (negative due to sign reversal), 0.92 and 0.78 for (cyanidin-3-galactoside), 1% to 2% (cyanidin-3-glucoside), 13% to PC1, 2, and 3, respectively. 17% (cyanidin-3-arabinose), 29% to 40% (peonidin-3-galactoside), Peonidin anthocyanin proportions signifi cantly increased with 3% to 5% (peonidin-3-glucoside) and 8% to 11% (peonidin-3-ara- later harvest dates for both years, particularly with the peonidin binoside) (Table 6). Cyanidin to peonidin anthocyanin ratios (cyd: arabinosides and galactosides (Fig. 2). Within harvest dates there pnd) varied from 0.8:1 (‘Ben Learʼ) to 1.5:1 (‘Early Blackʼ). The was a positive correlation between peonidin-3-galactosides with total anthocyanins, while cyanidin-3-galactosides were negatively Table 5. ANOVA (Type III sums of squares) for principal components correlated with total anthocyanins (data not shown). Proportions PC1, 2, and 3 from principal component analysis of anthocyanin of peonidin anthocyanins were signifi cantly positively correlated proportions from nine varieties as determined by SCAR fi ngerprint established in two cranberry beds. with total anthocyanins for the September harvest dates for 1998 (r = 0.14, P < 0.01) and 1999 (r = 0.32, P < 0.01). Mean square CLONES (VARIETIES) COMMON TO BOTH BEDS. Clones or variet- Source df PC1 PC2 PC3 ies were identifi ed to be present in both germplasm beds through Bed 1 14.9*** 35.9*** 4.0*** RAPD and SCAR fi ngerprinting technology (data not shown). Variety 8 9.3*** 5.7*** 0.1*** Bed and variety effects were highly signifi cant for PC1, 2 and 3 Bed × variety 6 1.3 0.4 0 (Table 5). Harvest date effects were highly signifi cant for PC1 and Harvest date (HD) 1 18.3*** 0.7 38.7*** 3. Interaction effects among the main effects were not signifi cant, Bed × HD 1 1.5 0.3 0 including variety by harvest date. Bed 1 exhibited a higher PC1 Variety × HD 8 3.2 0.2 0 value (1.5) than Bed 4 (0.0) indicating a higher proportion of the Error 32 1.3 0.3 0.01 peonidin anthocyanins. The later harvest date also exhibited higher ***Signifi cant at P < 0.001.

694 J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003.

9121-Genet 694 7/10/03, 10:59:17 AM Table 6. Means (across 2 years) of proportions of the six anthocyanins and absolute total anthocyanins for selected cranberry germplasm acces- sions and cultivars. Total Anthocyanin (%) anthocyanins Variety Cy-3-gal Cy-3-glc Cy-3-arab Pn-3-gal Pn-3-glc Pn-3-arab (µg·g–1 fruit) Common cultivars Stevens 33.3 1.4 14.8 35.9 3.4 11.0 352 Franklin 32.7 1.5 13.8 37.6 4.1 10.3 736 Early Black 39.9 1.9 16.0 30.9 3.4 8.0 665 Ben Lear 31.9 1.8 13.0 38.0 4.8 10.6 555 Accessions US89-7 52.1 3.7 20.7 16.2 2.8 4.6 328 US93-246 20.5 0.8 9.4 49.1 4.8 15.2 257 US92-27 27.5 1.2 12.6 41.8 4.5 12.1 921 US93-223 32.9 2.1 11.2 38.7 5.7 9.6 527 US89-10 27.4 0.7 17.1 38.1 2.0 14.7 253 US88-78z 34.3 1.7 12.5 40.3 4.3 8.9 422 US92-11z 34.4 1.6 20.0 29.5 3.3 11.1 558 SE 1.2 0.2 0.6 1.4 0.2 0.5 109 zData from 1998 only. most widely planted cultivar Stevens had about equal proportions, profi les in cranberry. 1:1 (1998) and 0.9:1 (1999), of cyanidin to peonidin anthocyanins The anthocyanin biochemical pathway, which shares early steps (Table 6). Accession US89-7 had the greatest proportion of cyanidin with the fl avonol and proanthocyanidin pathways, is a complex anthocyanins of 78% (1998), or a cyd:pnd ratio of 3.6:1. In 1999 the pathway consisting of numerous structural genes (Holton and proportion of cyanidin anthocyanins was slightly lower at 75% (3: Cornish, 1995). While enzymes early in the fl avonoid pathway, 1). Accession US93-246 (also accessions US93-247-248) exhibited (prior to chalcone isomerase), are thought to be encoded by large the low range (31%) of cyanidin anthocyanins or a ratio of 0.5: multigene families, it has been suggested that enzymes after the 1, cyd:pnd. Accession US93-246, however, was relatively low in chalcone synthase step are encoded by single active genes (Sparvoli overall anthocyanin content (Table 6). In contrast, the accession et al., 1994). However, genetic evidence suggests this may not be having the next lowest cyanidin proportion and relatively high the case (Jonsson et al., 1983). Dihydrofl avonol-4-reductase in anthocyanin production was US92-27, which had ≈43% cyanidin cranberry is encoded by a least two loci (Polashock et al., 2002). anthocyanins (Table 6). In maize, a number of regulatory genes controlling expression Proportions of arabinoside anthocyanins of the major cultivars of the anthocyanin pathway structural genes have been identifi ed range was narrow ranging from 24% to 26%. The range of arabi- (Dooner and Robbins, 1991). Boss et al. (1996) found almost all noside proportion across accessions was ≈21% (US93-223) to 32% the fl avonoid biosynthetic genes were expressed in many plant tis- (US89-10) (Table 6). Galactoside proportions of the major cultivars sues in grape, whereas, glucose-fl avonoid-3-O-glucosyltransferase, ranged narrowly between 69% to 71%. The highest proportion of the fi rst step from to anthocyanin production, was galactoside anthocyanins was ≈75% (US88-78), and lowest was only expressed in the fruit skin. A mutation in a regulatory gene ≈64% (US92-11) (Table 6). controlling the expression of a number of anthocyanin pathway genes including UDP-glucose:fl avonoid 3-O-glucotransferase was Discussion found in red-sports of white grapes (Kobayashi et al., 2001). The large variation (150-fold) in overall anthocyanin content Hong and Wrolstad (1986) reported cyanidin anthocyanin pro- in cranberry fruit suggests that considerable genetic variation is portions ranged from 53 % (1.1:1, cyanidin:peonidin) for the cultivar present for overall expression of the anthocyanin pathway. How- Howes to 65% (1.8:1, cyanidin:peonidin) for ‘Early Blackʼ. Sapers ever, variation in anthocyanin content has to be viewed in the and Hargrave (1987) reported cyanidin-3-galactoside percentages context of all the factors that can contribute to variation in whole ranging from 18.4% (cultivar not given) to 33.7% (‘Early Blackʼ). fruit samples. Anthocyanins are located predominately in the fruit The ‘Early Blackʼ accession in this study exhibited a higher per- skin in cranberry, appearing to be located in the outermost layer of centage (39% to 41%) of cyanidin-3-galactoside than Sapers and elongated epidermal cells and the inner layer of larger cells (Sapers Hargrave (1987). Genetic identity of cranberry varieties, however, et al., 1983a). Thus, variation in overall anthocyanin content of is problematic (Novy et al., 1994; Novy and Vorsa, 1995), and whole fruit samples is at least partially a function of fruit size, i.e. it is not certain whether the ‘Early Blackʼ clones between these amount of surface area, which is also a function of variety (Vorsa two studies are in fact identical genotypes. In addition, although and Welker, 1985). Recognized environmental effects which en- profi les were similar across the two years of this study, differences hance anthocyanin expression include water, light and temperature in growing year environments are also possible. The proportions stresses and nutrition. Furthermore, there is a negative correlation of cyanindin-3-galactoside, cyanindin-3-arabinoside, peonidin-3- between yield and many of the fl avonoid compounds, including galactoside, and peonidin-3-arabinoside for the cultivar Franklin, anthocyanins, proanthocyanidins, and total phenolics (Vorsa et al., however, appear to be similar between the studies with the exception 2000). Environmental stresses reduce fruit size, and thus yield. that peonidin anthocyanin proportions appear to be slightly higher Narrow sense heritability for total anthocyanins has been estimated in Sapers and Hargrave (1987). This study of a broader array of as high as h2 = 0.72 from parent-offspring regression suggesting germplasm identifi ed a broader range of variation of anthocyanin additive genetic effects contribute to major differences in overall

J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003. 695

9121-Genet 695 7/10/03, 10:59:21 AM anthocyanin production (Vorsa, unpublished data). by one locus may conjugate all three sugars, with allelic variation Analysis of the cranberry germplasm anthocyanin profi les yielding enzymes having different effi ciencies (Kms) for the various offers insight into the genetic and phenological effects that play sugars. Both genotypic and environmental (bed) effects impact (PC2 a role in the partitioning of fl avonoid resources. Contrary to the and PC3) glycosylation profi les. Phenological effects were only conclusions of Sapers and Hargrave (1987), this study suggests signifi cant for galactoside versus glucoside/arabinoside trade-offs there is signifi cant genotypic variation for individual anthocyanin (PC3), suggesting that galactosyltransferase expression may not proportions. Evidence for both qualitative and quantitative variation be coupled with glucosylation/arabinosylation. Since glucose and for proportions of the aglycone (in the conjugated form), cyanidin arabinose are six and fi ve carbon sugars, respectively, with identical versus peonidin, was found. Accession US89-7 producing ≈80% stereochemistry for the fi ve hydroxyl groups, it could be reasoned cyanidin anthocyanins is an outlier (Fig. 1) suggesting possible that both sugars are conjugated by one glycosyltransferase enzyme homozygosity for a defective structural methyltransferase gene or a in cranberry. The signifi cant trade-offs between glucosides versus regulatory gene of methyltransferase. Jonsson et al. (1983) suggested arabinosides (PC2) may refl ect allelic differences in nucleotide at least two anthocyanin methyltransferases were present in Petunia affi nities or carbohydrate availability. hybrida under the control of four regulatory genes. If more than In conclusion, the signifi cant genetic variation for cranberry one methyltransferase locus exists in cranberry, then the simplest anthocyanin profi les provides opportunities to breed for specifi c explanation would be that a regulatory gene is responsible for the anthocyanin profi les. The greatest opportunity appears to be for relatively low peonidin content in US89-7. Signifi cant quantitative proportion of cyanidin versus peonidin anthocyanins, with par- variation is also present, suggesting more subtle allelic differences ticularly high (80%) cyanidin profi les the most readily achievable for structural and/or regulatory genes. since it may be a qualitative trait. Cranberry varieties with not only Grisebach (1982) suggests that methylation occurs at the cin- high overall anthocyanin levels, but with increased proportions of namic acid stage or later at the chalcone-fl avonone stage. The greater cyanidin anthocyanins would be expected to have higher antioxidant proportions of peonidin in later harvests and positive correlation capacity potential due to the apparently higher antioxidant capacity with overall anthocyanin content indicates that the methylation of cyanidin relative to peonidin that was reported by Wang et al. step in cranberry occurs post-glycosylation by methyltransferases (1997). Although apparent genetic variation exists for glycosylation as suggested by Holton and Cornish (1995). Do et al. (1995) found profi les, the variation in the current germplasm collection may not the Vitis vinifera 3-O-glucosyltransferase to have the highest ac- be practically signifi cant. tivity with cyanidin, and considerably reduced activity with both Since cranberry is an asexually propagated crop, the lack of 3ʼ and 5ʼ methylated , peonidin and malvidin. The numerous sexual generations, reduces the opportunity of exposing positive correlation of peonidin anthocyanin proportion with total mutations as homozygous recessive traits. Screening undomesti- anthocyanins and later harvest dates suggests that the expression of cated germplasm may provide additional unique profi les and genes methyltransferase(s) is not or only partially coupled with expres- for breeding. However, opportunities should exist with inbreeding sion of earlier pathway genes. If methylation occurs early in the for identifying variant regulatory or structural alleles for glycosyl- pathway, i.e., at chalcone isomerase or synthase steps, then one ation and other fl avonoid pathway genes. would not expect to have phenological effects infl uencing propor- tions of cyanindin versus peonidin proportions. Literature Cited Apparent genetic variation was found for glycosylation profi les. Boss, P.K., C. Davies, and S.P. Robinson. 1996. Expression of anthocy- The proportions of arabinosides versus glucosides (PC2) accounted anin biosynthesis pathway genes in red and white grapes. Plant Mol. for a major portion of the variation in glycosylation, followed by Biol. 32:565–569. arabinosides/glucosides versus galactosides (PC3). In many species, Cao, G., E. Sofi c, and R.L. Prior. 1996. Antioxidant and pro-oxidant anthocyanidins are conjugated to glucose by UDP glucose:fl avo- behavior of fl avonoids: Structure-activity relationships. Free Radical noid 3-O-glucosyltransferase to form anthocyanidin 3-glucosides Biol. Med. 22:749–760. followed by further modifi cation to other glycosides (Holton and Do, C.B., F. Cormier, and Y. Nicolas. 1995. Isolation and characterization Cornish, 1995). Glycosyltransferases represent a large supergene of a UDP-glucose:cyanidin 3-O-glucosyltransferase from grape cell family and appear to have fairly broad affi nity for substrates, but suspension cultures (Vitis vinifera L.). Plant Sci. 112:43–51. exert regioselectivity and regiospecifi city for the sugar acceptor Dooner, H.K. and T.P. Robbins. 1991. Genetic and developmental control (Vogt and Jones, 2000). Multiple glucosyltransferases have been of anthocyanin biosynthesis. Annu. Rev. Genet. 25:173–199. Ford, C.M., P.K. Boss, and P. B. Hoj. 1998. Cloning and characterization identifi ed in Petunia hybrida with specifi city for both, the substrate of Vitis vinifera UDP-glucose: Flavonoid 3-O-glucosyltransferase, a and sugar donor, as well as differential expression in various tissues. homologue of the enzyme encoded by the maize bronze-1 locus that An UDP-glucose:fl avonoid 3-O-glucosyltransferase isolated from may primarily serve to glucosylate anthocyanindins in vivo. J. Biol. Vitis vinifera did not exhibit activity with UDP-galactose (Ford et Chem. 273:9224–9233. al., 1998). Similarly, evidence suggests that in some species such Gee, J.M. M.S. DuPont, A.J. Day, G.W. Plumb, G. Williamson, and I.T. as Eustoma grandifl orum, which produces predominantly 3-O- Johnson. 2000. Uptake of quercetin glycosides involves deglycosylation galactosylated fl avonols, that 3-O-glucosylated fl avonols can be and interaction with sodium-dependent glucose transport pathway in increased substantially with transformation and expression of a rat small intestine. J. Nutr. 130:2765–2771. UDP-glucose:fl avonoid-3-O-glucosyltransferase (Schwinn et al., Grisebach, H. 1982. Biosynthesis of anthocyanins, p. 69–92. In: P. Marka- 1997). It has been suggested that the supply and quality of nucleo- kis (ed.). Anthocyanins as food colors. Academic Press, New York. Hollman., P.C.H. and M.B. Katan. 1997. Absorption, metabolism, and tides would be conserved during evolution leading to high substrate health effects of dietary fl avanoids in man. Biomedicine Pharmaco- specifi city for the sugar donor (Vogt and Jones, 2000). Thus, it could therapy 51:305–310. be postulated that cranberry has up to three glycosyltransferase Hollman, P.C.H., J.M.P. van Trijp, M.N.C.P. Buysman, M.S. v.d. Gaag, enzymes including galactosyltransferase, arabinosyltransferase M.J.B. Memgelers, J.H.M. de Vries, and M.B. Katan. 1997. Relative and glucosyltransferase, the genes for which would be located bioavailability of the antioxidant fl avonoid quercetin from various at respective loci. Alternatively, one glycosyltransferase encoded foods. FEBS Lett. 418:152–156.

696 J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003.

9121-Genet 696 7/10/03, 10:59:26 AM Holton, T.A. and E.C. Cornish. 1995. Genetics and biochemistry of Hort. Sci. 108:246–249. anthocyanin biosynthesis. Plant Cell 7:1071-1083. Sapers, G.M., J.G. Phillips, H.M. Rudolf, and A.M. DiVito. 1983b. Hong, V. and R.E. Wrolstad. 1986. Cranberry juice composition. J. Assn. Cranberry quality: Selection procedures for breeding programs. J. Off. Anal. Chem. 69:199–207. Amer. Soc. Hort. Sci. 108:241–246. Hong, V. and R.E. Wrolstad. 1990. Use of HPLC separation/photodiode Satue-Gracia, M.T., M. Haeinonen, and E.N. Frankel. 1997. Anthocyanins array detection for characterization of anthocyanins. J. Agr. Food Chem. as antioxidants on human low-density lipoprotein and lecithin-liposome 38:708–715. systems. J. Agr. Food Chem. 45:3362–3367. Jonsson, L.M.V., P. de Vlaming, H. Wiering, M.E.G. Aarsmanard and Schwinn, K.E., K.M. Davies, S.C. Deroles, K.R. Markham, R.M. Miller, A.W. Schram. 1983. Genetic control of anthocyanin-O-methyltrans- J.M. Bradley, D.G. Manson, and N.K. Given. 1997. Expression of an ferase activity in fl owers of Petunia hybrida. Theor. Appl. Genet. 66: Antirrhinum majus UDP-glucose: Flavonoid-3-O-glucosytransferase 349–355. transgene alters fl avonoid glycosylation and acylation in lisianthus Kobayashi, S., M. Ishimaru, C.K. Ding, H. Yakushiji, and N. Goto. (Eustoma grandifl orum Grise.). Plant Sci. 125:53–61. 2001. Comparison of UDP-glucose: Flavonoid 3-O-glucosyltransferase Setchell, K.D.R., N.M. Brown, P. Desai, L. Zimmer-Nechemias, B.E. (UFGT) gene sequences between white grapes (Vitis vinifera) and their Wolfe, W.T. Brashear, A.S. Kirschner, A. Cassidy, and J.E. Heubi. 2001. sports with red skin. Plant Sci. 160:543–550. Bioavailability of pure isofl avones in healthy humans and analysis of Lamuela-Raventos, R.M. and A.L. Waterhouse, 1994. A direct HPLC commercial soy isofl avone supplements. J. Nutr. 131:1362S–1375S. separation of wine phenolics. Amer. J. Enol. Viticult. 45:1–5. Sparvoli, F.C., C. Martin, A. Scienza, G. Gavazzi and C. Tonelli. 1994. Mizuma, T., K. Ohta, and S. Awazu. 1994. The β-anomeric and glucose Cloning and molecular analysis of structural genes involved in fl a- preferences of glucose transport carrier for intestinal active aborption of vonoid and stilbene biosynthesis in grape (Vitis vinifera) Plant Mol. monosaccharide conjugates. Biochim. Biophys. Acta 1200:117–122. Biol. 24:743–755. Novy, R.G., C. Kobak, J. Goffreda, and N. Vorsa. 1994. RAPDs identify Vogt, T. and P. Jones. 2000. Glycosyltransferases in plant natural product varietal misclassifi cation and regional divergence in cranberry (Vac- synthesis:characterizarion of a supergene family. Trends Plant Sci. 5: cinium macrocarpon Ait.). Theor. Appl. Genet. 88:1004–1010. 380–386. Novy, R.G. and N. Vorsa. 1995. Identifi cation of intracultivar genetic Vorsa, N. and W.V. Welker. 1985. Relationship between fruit size heterogeneity in cranberry using silver-stained RAPDs. HortScience and extractable anthocyanin content in cranberry. HortScience 20: 30:600–604. 402–403. Polashock, J.M., R.J. Griesbach, R.F. Sullivan, and N. Vorsa. 2002. Vorsa, N., J. Polashock, A. Howell, D. Cunningham, and R. Roderick. Cloning of a cDNA encoding the cranberry dihydrofl avonol-4-re- 2000. Evaluation of fruit chemistry in cranberry germplasm: Potential ductase (DFR) and expression in transgenic tobacco. Plant Sci. 163: for breeding varieties with enhanced health constituents. 7th Intl. Symp. 241–251. Vaccinium Culture, termas de Chillán, Chillán, Chile, 4–9 Dec. Sapers, G.M. and D.L. Hargrave. 1987. Proportions of individual an- Wang, H. G. Cao, R.L. Prior. 1997. Oxygen radical absorbing capacity thocyanins in fruits of cranberry cultivars. J. Amer. Soc. Hort. Sci. of anthocyanins. J. Agr. Food Chem. 45:304-309. 112:100–104. Wrolstad, R.E., J.D. Wightman, and R.W. Durst. 1994. Glycosidase Sapers, G.M., S.B. Jones, and G.T. Maher. 1983a. Factors affecting activity of enzyme preparations used in fruit juice processing. Food the recovery of juice and anthocyanin from cranberies. J. Amer. Soc. Technol. (Nov.):90, 92–94, 96, 98.

J. AMER. SOC. HORT. SCI. 128(5):691–697. 2003. 697

9121-Genet 697 7/10/03, 10:59:31 AM